Research On Key Technologies Of Automatic Question Answering System In Judicial Field Based On Knowledge Graph

With the advancement of the work of governing the country according to law,the work of law popularization is receiving more and more attention.Compared with traditional domain databases,domain knowledge graphs can effectively integrate and associate massive text data,and in-depth mining of semantic relationships in text data.At the same time,the question and answer technology based on knowledge graphs is conducive to quickly and accurately obtaining effective information.How to combine new technologies to make law popularization work intelligent and convenient is a problem worthy of study.In order to solve this problem,this paper mainly studies the key technology of automatic question answering system based on the knowledge map of the judicial field.The main work results are as follows:(1)Propose a named entity recognition model for legal text information.Aiming at the characteristics of legal texts,a Bi LSTM-CRF-LER(Bi-directional Long Short-Term Memory-Conditional Random Field-Judicial Entity recognition)model is proposed to extract legal entities at the character level,using word vector technology and a combination of two-way long and short-term memory models.Conditional random field model to obtain the optimal sequence labeling.Through comparative experiments with several other entity recognition models in related researches in the judicial field,the correctness and advantages of this model are proved.(2)A rule-based method for extracting legal text relations is proposed.Based on the results of entity recognition,the entity user dictionary in the judicial field is constructed,and rules are gradually formulated through the results of word segmentation and part-of-speech tagging,and the final correct part-of-speech tagging results are obtained.In addition,a comparative experiment was set up to prove the rationality and accuracy of this method.(3)Construct a knowledge map in the judicial field.According to the results of entity relationship extraction,this paper constructs a knowledge map in the judicial field,and at the same time completes the data operation of the knowledge map based on the Neo4 j graph database.By correctly constructing a knowledge graph in the judicial field,the correctness of the entity and relationship extraction method in this paper is verified.(4)Realize an automatic question and answer system based on the knowledge map of the judicial field.Based on the construction of the judicial domain knowledge graph,through question analysis and entity relationship links,the corresponding query results in the knowledge graph are finally optimized into the corresponding answer and then the user answer is returned,and the accuracy experiment is set to test the accuracy of the system.The judicial field corpus database constructed using this article can solve the current shortage of public data sets in the judicial field to a certain extent.The entity relationship extraction method proposed in this paper can ensure the efficient extraction of entities and relationships in legal text information.The knowledge graph and automatic question answering system in the judicial field constructed by this article can provide reference for the research of Chinese knowledge graph construction technology and automatic question answering technology in other fields.